Mikkel Lykkegaard

Mikkel’s research is concerned with Bayesian inversion for hydrogeology – specifically using cutting-edge Markov Chain Monte Carlo (MCMC) techniques for groundwater flow parameter estimation and uncertainty quantification. He has contributed to the Multilevel Delayed Acceptance (MLDA) sampler to PyMC3 in collaboration with researchers from the ALan Turing Institute, and written an open source gradient-free Delayed Acceptance MCMC framework, tinyDA, which he uses in his own research.

His current research is concerned with adaptive optimal design of groundwater surveys, motivated by the question "where to drill next?, where he is investigating various adjoint methods in conjunction with Monte Carlo uncertainty quantification. He is also working on different parallelisation strategies for Delayed Acceptance MCMC.

Research Interests:

• Environmental (geo-)hydrology and hydroinformatics
• Monte Carlo methods
• Uncertainty quantification for Bayesian inverse problems
• Distributed environmental models
• Environmental fate and risk assessment of pollutants